Abstract
The rhizosphere has been called "one of the most complex ecosystems on earth" because it is a hotspot for interactions among millions of microbial cells. Many of these are microbes are also participating in a dynamic interplay with host plant tissues, signaling pathways, and metabolites. Historically, breeders have employed a plant-centric perspective when trying to harness the potential of microbiome-derived benefits to improve productivity and resilience of economically important plants. This is potentially problematic because: (i) the evolution of the microbes themselves is often ignored, and (ii) it assumes that the fitness of interacting plants and microbes is strictly aligned. In contrast, a microbe-centric perspective recognizes that putatively beneficial microbes are still under selection to increase their own fitness, even if there are costs to the host. This can lead to the evolution of sophisticated, potentially subtle, ways for microbes to manipulate the phenotype of their hosts, as well as other microbes in the rhizosphere. We illustrate this idea with a review of cases where rhizosphere microbes have been demonstrated to directly manipulate host root growth, architecture and exudation, host nutrient uptake systems, and host immunity and defense. We also discuss indirect effects, whereby fitness outcomes for the plant are a consequence of ecological interactions between rhizosphere microbes. If these consequences are positive for the plant, they can potentially be misconstrued as traits that have evolved to promote host growth, even if they are a result of selection for unrelated functions. The ubiquity of both direct microbial manipulation of hosts and context-dependent, variable indirect effects leads us to argue that an evolutionary perspective on rhizosphere microbial ecology will become increasingly important as we continue to engineer microbial communities for crop production.